What problem does it solve? Managing Docker stacks across two remote hosts (LinuxBro and SuperBro) requires knowing which MCP server, environment ID, compose path, and redeploy method applies to each host, plus avoiding dangerous mistakes like wiping Traefik secrets or leaving containers stuck in Created state. This Skill encodes all of that operational knowledge so stack operations are routed and verified correctly. ## Core Features & Use Cases - Host-aware MCP routing: Selects the correct Dockhand MCP server (dockhand-linuxbro or dockhand-superbro), Tailscale endpoint, and environment ID based on the host named in the request. - Stack lifecycle operations: Lists stacks, reads and updates compose files, redeploys or restarts stacks and individual containers, edits container files, and reads logs through Dockhand MCP tools. - Traefik and infrastructure safeguards: Documents secret requirements (CF_DNS_API_TOKEN in stack variables), tunnel route prerequisites, and mandatory post-deploy verification including stuck-container sweeps. - Use Case: Ask to redeploy the immich stack on LinuxBro; the Skill routes to the correct MCP server, applies the compose update with restart, then verifies no containers are stuck and routes still respond. ## Quick Start Use the dockhand skill to restart the media-core stack on LinuxBro and confirm all containers are running afterward.